1. Field of the Invention
The present invention relates to a semiconductor device that uses a semiconductor film having crystallinity and a manufacturing method of such a semiconductor device. In particular, the invention relates to a semiconductor device having a CMOS structure that uses a silicon film as a semiconductor film.
2. Description of the Related Art
In recent years, the CMOS technology using insulated-gate transistors has been developed extensively. However, as described in Japanese Unexamined Patent Publication Nos. Hei. 4-206971 and Hei. 4-286339, transistors using a crystalline silicon film as the active layer have such a tendency that the electrical characteristic is shifted to the depletion direction (negative side) in n-type transistors and to the enhancement direction (negative side) in p-type transistors. The reason for this tendency is considered the difference in work function between the gate electrode and the active layer that depends on the conductivity type.
FIG. 2 schematically shows the above-mentioned electrical characteristics (Id-Vg characteristics) of transistors. The horizontal axis represents the gate voltage Vg and the vertical axis represents the vertical axis represents the drain current Id. Reference numerals 201 and 202 denote Id-Vg characteristics of an n-type transistor and a p-type transistor, respectively. Intersections of the horizontal axis and the Id-Vg characteristics 201 and 202 indicate threshold voltages.
Reference numeral 203 denotes a window width Vwin which is defined as a difference between a threshold voltage V.sub.th, n of the n-type transistor and a threshold voltage V.sub.th, p of the p-type transistor, i.e., V.sub.th, n -V.sub.th, p. Further, reference numeral 204 denotes a window center Vcen which is defined as a value at the center of the window, i.e., (1/2)(V.sub.th, n +V.sub.th, p).
In conventional CMOS circuits, since the window width Vwin is shifted to the negative side, the window center Vcen is smaller than 0 V. The above-mentioned publication No. Hei. 4-206971 points out that deviations of output voltages due to the difference between the threshold voltages cause deterioration in the characteristics of a CMOS circuit.
One method for solving this problem is to control the threshold voltages by adding, to the channel forming regions, an impurity (phosphorus or boron) that imparts one conductivity type (hereinafter called "channel doping method"). However, this method has a problem that impurity ions cause carrier scattering, possibly reducing the operation speed. In particular, in the deep submicron range in which the channel length is as short as 0.01-0.1 .mu.m, only one to several impurity ions exist in the channel region. There is a report that the existence of impurity ions drastically changes the electrical characteristics.
It is now necessary to refer to a short channel effect preventing technique (pinning technique) that is proposed by the present inventors. This technique will be outlined below with reference to FIGS. 3A-3C. Although FIGS. 3A-3C show a thin-film transistor formed on an insulative substrate, the same things apply to a transistor formed within a semiconductor substrate.
The short channel effect is a generic term representing such phenomena as a reduction in breakdown voltage due to the punch-through phenomenon and deterioration of the subthreshold characteristic. These phenomena are caused such that the drain-side depletion layer expands to the source region to establish a situation that carriers cannot be controlled only by the gate voltage.
The pinning technique is a technique for preventing the expansion of the drain-side depletion layer by providing, artificially and locally, impurity regions in the channel forming region. The inventors use the term "pinning" as meaning "preventing."
Specifically, the active layer of a transistor is configured as shown in FIGS. 3A-3C. In FIG. 3A, reference numerals 301-303 denote a source region, a drain region, and a channel forming region, respectively. Impurity regions 304 are artificially formed in the channel forming region 303. In the channel forming region 303, regions 305 other than the impurity regions 304 are substantially intrinsic regions where carriers are allowed to move. Symbols L and W denote a channel length and a channel width, respectively.
The impurity regions 304 are obtained by forming a fine pattern by an electron beam lithography method or the like. Although FIG. 3A shows an example in which the impurity regions 304 are formed in a linear pattern, they may be formed in a dot pattern.
FIG. 3B is a sectional view taken along line A-A' in FIG. 3A.
Reference numeral 306 denotes a substrate having an insulative surface.
FIG. 3C is a sectional view taken along line B-B' in FIG. 3A.
The impurity regions 304, which are disposed in the channel forming region 303, form regions (energy barriers) where the diffusion potential is locally high in the channel forming region 303. The energy barriers can effectively prevent (i.e., pin) the drain-side depletion layer from expanding toward the source side.
Sufficiently high energy barriers can be formed by adding any of oxygen, nitrogen, and carbon. B (boron) and P (phosphorus) may be added in the cases of an n-type transistor and a p-type transistor, respectively.
With the above structure, it is expected that a reduction in threshold voltage as one aspect of the short channel effect can be prevented effectively. Naturally, it is also possible to prevent a reduction in breakdown voltage due to the punch-through phenomenon and deterioration of the subthreshold characteristic.
It is expected that the structure of FIGS. 3A-3C causes a narrow channel effect separately from the above effects. That is, a narrow channel effect can be caused artificially in the carrier movement regions 305 by sufficiently narrowing the intervals between the impurity regions 304.
As described above, the pinning technique that is proposed by the inventors is effective in such a range as covers device elements in which the short channel effect occur (channel length: 2 .mu.m or less) and finer device elements in the deep submicron range (channel length: 0.01-0.1 .mu.m).
However, the shift of the window center Vcen due to the difference in work function between the gate electrode and the active layer that was described above in the conventional example (FIG. 2) similarly occurs even if the pinning technique is employed. Therefore, in the submicron range, it is necessary to control the threshold voltages while preventing the short channel effect.